Abnormal subthalamic nucleus functional connectivity and machine learning classification in Parkinson's disease: a...

Area of Science:

• Neuroscience
• Radiology
• Biomarkers

Background:

• Parkinson's disease (PD) is a progressive neurodegenerative disorder impacting motor function due to aberrant basal ganglia activity.
• Subthalamic nucleus (STN) dysfunction is implicated in PD's motor symptoms.
• Resting-state functional magnetic resonance imaging (rs-fMRI) offers insights into brain network alterations.

Purpose of the Study:

• To characterize STN functional connectivity (FC) abnormalities in PD patients.
• To evaluate the utility of STN-FC patterns as machine learning (ML) biomarkers for PD classification.

Main Methods:

• Analysis of pooled rs-fMRI data from 232 participants (158 PD patients, 74 healthy controls [HCs]) across multiple repositories.
• Seed-based FC analysis focusing on bilateral STNs, with group comparisons using t-tests and Gaussian Random Field (GRF) correction.
• Development of a support vector machine (SVM) classifier using significant FC features for diagnostic classification.

Main Results:

• PD patients exhibited significant bilateral reductions in STN FC compared to HCs.
• Specific decreased connectivity was observed between the STN and temporal/parietal regions (e.g., superior temporal gyrus, supramarginal gyrus).
• The SVM classifier achieved high diagnostic accuracy (89.1%), sensitivity (97.7%), specificity (75.8%), and an AUC of 0.931.

Conclusions:

• STN-temporal/parietal hypoconnectivity may represent a core feature of PD.
• STN-centric FC patterns hold significant translational potential as diagnostic biomarkers for PD.
• These findings support improved PD classification and personalized neuromodulation strategies.